SUMMARY Over 30% of all proteins bind metal ions, including transition metals, for structural and functional purposes. Even though there are rich experimental structural and thermodynamic information on protein-ion complexes, our understanding of the physical basis for the specificity and selectivity in protein-ion recognition remains lacking. There is a great need for accurate physical models and efficient simulation software to enable computational study of protein-ion systems. We propose to develop a next generation classical force field AMOEBA+, based on the existing AMOEBA potential, to systematically model permanent electrostatics, repulsion, dispersion, charge penetration, polarization, charge transfer, and ligand field effect after quantum mechanical energy decomposition and experimental data. The new potential and high- performance molecular simulation software (Tinker for CPU & GPU systems) will allow us to comprehend the structural, physical and thermodynamic driving forces underlying protein-ion recognition using molecular dynamics simulations. Given the fundamental importance of protein interaction with transition metals including Zn, Cu, Ni, Co, Fe, and Mn, this research will have a broad impact on advancing our scientific knowledge about ions in biomolecular structure and function, and lead to new computational tools to accelerate the design of new diagnostic and therapeutic molecules targeting protein-ion interactions.